In the information storage industry, increases in the data capacity of tape have been achieved with thinner tape substrates and with various data compression techniques. Advances in the magnetic tape media and tape head technologies have generated further increases in data capacity by increasing both the number of tracks on a magnetic tape (by narrowing the width of each track) and the number of data read/write "gaps" or elements on the head. For example, the IBM 3490-E magnetic tape subsystem for 12.7 mm (1/2 inch) wide tape employs a head with 36 data read/write elements and has a tape capacity of about 800 megabytes (MB). 3490-E tape drives perform serpentine (bi-directional) linear (as opposed to helical scan) recording and, to reduce the number of tape rewinds and thereby improve performance, sets of data head elements are interleaved. In an interleaved head, element pairs having a read-element/write-element configuration (when viewed toward the face of the head) alternate with element pairs having an opposite, write-element/read-element configuration, each pair of elements being associated with one track on the tape. When the tape travels in one direction, the element pairs having one configuration access the associated tracks (such as the even numbered tracks) in a read-after-write manner and when the tape travels in the opposite direction, the element pairs having the other configuration access the other associated tracks (the odd numbered tracks), also in a read-after-write manner. To further enhance performance with such a large number of tracks requiring an equally large number of closely spaced elements, a magneto-resistive transducer formed through thin film deposition techniques can be employed as the tape head.
However, the number of parallel, longitudinal tracks which can be established on tape media has been limited by the number of read/write elements which could be fabricated on a head to write/read narrower tracks. Therefore, data tape drives have been designed to process a tape using a head having fewer sets of read/write elements than there are tracks on the tape. The tracks are divided into groups, each containing the same number of tracks as there are read/write element pairs in the head. To access all of the groups, the head is indexed transversely relative to the tape width, such as with a stepper motor or voice coil driven springs, into a number of discrete positions corresponding to the number of groups of tracks. For example, a head having eight read/write pairs can accommodate a tape having 24 tracks if the tracks are divided into three groups of eight tracks each and the head has three index positions. To reduce the distance the head travels between index positions, the groups are preferably interleaved. In the preceding example, each of three groups contained eight tracks. If tracks are numbered consecutively (0-23), group interleaving results in tracks 0, 3, 6, . . . , and 21 being in the first group and accessible by the head being indexed to position 0. Similarly, tracks 1, 4, 7, . . . and 22 are in the second group, accessible in index position 1, while tracks 2, 5, 8, 23 are in the third group, accessible in index position 2. Serpentine recording techniques can also be employed to improve access efficiency.
Despite such advances in data capacity, still further increases are desired, such as would be possible with a 12.7 mm wide tape having 64 or even 128 tracks. However, even when a head is indexed, there is a practical limit to the ability of a multi-track head to accurately and reliably record data to and read data from a tape having such a large number of very narrow tracks. Problems can be caused by track misregistrations, such as tape edge variations, environmental thermal expansion and contraction and inaccuracies in the path the tape follows in a drive, by inaccuracies in the formatting of tracks on the tape itself and by dimensional and spacing deviations during the manufacturing of the head. It can be appreciated that even a minute "wobble" in the tape or a misalignment in the head can result in significant signal degradation, such as crosstalk and dropout, if a 12.7 mm tape has 128 tracks, each with a width of about 80 microns.
Consequently, a tape head actuator has been developed which is capable of indexing a tape head to one of several positions during track seek operations. For example, to access a tape having 128 data tracks, a head having thirty-two read/write elements on the head indexes among four positions. Moreover, the head actuator is also capable of rapidly adjusting the position of the head under servo control to precisely follow a set of tracks during read and write operations. In a drive employing such actuator, the tape head has servo read elements or gaps for reading servo signals previously recorded onto one or more specially recorded servo tracks. Each servo element generates a position error signal (PES) which is employed by a position servo loop to determine the transverse position of the servo elements relative to the servo tracks. The loop then transmits a signal to the head actuator to rapidly move the head by very small amounts as necessary to enable precise track following.
To improve the accuracy with which the servo loop operates, the tape can have two or more servo areas, comprising a set of one or more servo tracks, spaced across the width of the tape; the tape head has a corresponding number of sets of servo elements. The PES's generated by the servo elements are concurrently read and averaged. The head position is maintained by the servo loop in response to the average, rather than the PES from any one servo element. Such redundancy makes the servo loop less susceptible to error or failure due to an error or failure by any one servo element. One such system includes three symmetrically spaced servo areas, each having three adjacent servo tracks, parallel to the data tracks. In each servo area, servo signals are recorded on the outer two (upper and lower) servo tracks at one frequency while servo signals are recorded on the middle servo track at a different frequency.
To provide for four head index positions, each set of servo elements has transversely spaced upper and lower servo elements, one of which is active at one time. During indexing, either the upper or lower servo element of each set of servo elements is positioned opposite the upper or lower boundary or edge between the middle servo track and the upper or lower servo track. Then, during data access, the servo loop attempts to maintain the magnetic center of the active servo elements in alignment with the selected upper or lower edge in the respective servo areas. It will be appreciated that the combination of two groups of servo elements (upper and lower) and two servo edges (upper and lower) makes the four index positions available and that other combinations provide other numbers of index positions.
If all of the servo elements are perfectly formed on the tape head, both in size and in position relative to each other servo element, all active servo elements will be maintained centered over the desired servo area edge within the accuracy of the servo control system and the average of the PES's generated by the active servo elements should be substantially the same as the PES of any individual PES. The advantage of employing redundant servo areas is apparent if one of the servo elements becomes disabled: accurate servo positioning can still be maintained. However, such manufacturing precision cannot be economically achieved in commercial quantities and the servo elements can have slight dimensional and spacing deviations whereby their magnetic centers are not perfectly aligned with each other or with the servo areas of the tape. Consequently, when one servo element is centered on the desired servo edge, the other servo elements in the group will likely not be centered. Consequently, use of an average PES may not result in any servo element being centered on the servo area edge, leading to potential mistracking and read and write errors, such as overwriting onto adjacent data tracks. Discrepancies of this nature are referred to as element-to-element offsets.
As previously noted, data is accessed (written, read or erased) by element pairs comprising spaced apart read and write elements. One type of tape head is formed with left and right modules on which are located the data elements; the read element of one pair is formed on one module and the associated write element is formed on the other module. The servo elements on such a tape head are also paired, left and right. The right servo elements are enabled when the tape is travelling in a forward direction; the left servo elements are enabled when the tape is travelling in the reverse or backward direction. However, because the two modules may not be perfectly manufactured or bonded to each other, module-to-module discrepancies or offsets can occur, resulting in mistracking and additional read and write errors. Moreover, the head itself may be slightly angularly misaligned, leading to an azimuthal offset. Finally, because the servo areas of each tape may not be perfectly aligned and recorded, they may vary from one tape to the next. Therefore, different combinations of tape and tape drives can result in different degrees of read and write errors from the aforementioned discrepancies.